About Best in IT: Practical AI, Automation and Developer Tools

Greetings from Best in IT, authored by technology aficionado Artur Poniedziałek. This digital platform serves as a repository for insights gathered while exploring AI, automation, and developer tools in practical IT scenarios. Initially established to bridge the gap between technological advancements and their real-world applicability, Best in IT tackles the challenge of translating exciting tech news into actionable solutions.

Artur's work predominantly revolves around local AI models and private AI assistants, with recent emphasis on leveraging these technologies on local hardware. His research interests include the trade-offs between model size, speed, memory, and quality when running models locally. He is keen on exploring workflows where AI can enhance productivity without compromising control over sensitive data.

Key focus areas for Artur include:

- Local Large Language Models (LLMs) and inference

- Retrieval Augmented Generation (RAG) systems, vector databases, chunking, and retrieval

- AI agents, orchestration, and human approval workflows

- Python automation and developer tools

- Local image generation using ComfyUI

- Speech-to-text conversions using Whisper and video transcription

- UI testing with Playwright and vision-language models

Artur's exploration of local AI highlights the benefits of privacy, predictable costs, lower latency, and control over data handling. However, he acknowledges the challenges associated with running models locally, such as GPU memory limits, driver issues, model formats, hardware compatibility, and the gap between theoretical performance and practical utility.

To bridge these gaps, he conducts extensive testing on Windows and AMD hardware, including specific applications like whisper.cpp with Vulkan for transcription, ComfyUI through its API for automated image generation, RAG chunking strategies, and integration of Playwright screenshots with local vision models for UI analysis.

While Best in IT does not claim any singular tool as superior, it serves as a comprehensive documentation of what has worked, what has failed, what required adjustments, and what modifications Artur would implement in future endeavors. The blog features a mix of hands-on tutorials, benchmarks, comparisons, and candid notes from projects that faced unforeseen challenges. Artur believes that these candid accounts of setbacks are as valuable as successful implementations, offering a well-rounded perspective for readers.

Best in IT caters to a diverse audience, including developers, IT professionals, technical managers, consultants, and curious builders who seek practical insights into AI and software tools. There is no requirement for an AI researcher background; anyone looking to understand whether local models are worth the effort, how to construct more reliable RAG systems, how to automate repetitive tasks with Python, or how to responsibly integrate AI into business environments will find valuable content here.

Artur pledges to share his testing experiences, explain the rationale behind his decisions, and continually refine the content as the landscape of AI and technology evolves.